Rapid Prototyping

I started a different post last weekend, but all I ever did was get the pictures together and it’ll take a bit more work to finish up the other content I was planning for that post, so here’s a completely different topic.

One of the nice things about having a 3d printer is the ability to design something and have a physical copy of it a short time later (as short as minutes to hours, depending on size). This allows for rapid iteration and fine tuning of designs, and I believe is typically what 3d printers are used for commercially (excluding companies that print as a service, like Shapeways). Here’s the last two weeks worth of work on one of my projects:

A 3d printed glove was the original goal, but it quickly became obvious that I was basically designing armor so now that’s the plan.

Version 1 (white ABS)

The white pieces make up the first set of iterations. They were all printed in a weekend, and the takeaway from that work was the inner shape including cutouts on the bottom of the finger to allow the armor to be tightly fitted but still let the finger flex. I ended that set by putting a hinge between the two joint segments that I’d printed, at which point I realized that my finger was just crooked enough that a straight hinge won’t work if I’m trying to have it fit snugly. Also, printing on end not only won’t work for putting hinges on both ends, but it puts the strength in the wrong direction for making a strong hinge.

The main advantages of printing on end was that I could minimize the plastic used in each piece by printing exactly the shape of my finger, and the prints were fast: if the printer was warmed up I could go from exporting the design to holding the printed piece in about ten minutes (though I was over-eager once and tried to remove it from the printer while it was still hot enough to be soft; completely bent that piece).

Version 2 (black PLA, blue ABS)

The blue and black pieces were printed over the span of about two weeks after the initial set. I forget why I initially tried printing in PLA, but it was too stiff to get the hinges to snap together without breaking the tabs off.

This set was nearly a complete rework of the original script to remove complexity. On the first set I’d just kept adding more and more operations to refine the shape to the point where it wouldn’t render quickly in OpenSCAD, so I basically redesigned the hinge to consist of two pieces per side and modeled the new finger-top geometry as a single piece.

The reworked design is printed with the top of the finger flat on the bed, and it’s more angular instead of rounded to match the finger. That makes it printable, but the hinges needed support (that kept breaking off, which is why they look kind of messy), it took more plastic per print, and the print times went up to I think around fifteen to twenty minutes per segment. I also tried out putting the hinge at an angle (the connected black and blue pair), and while that fundamentally worked it didn’t match my finger at all.

The resulting piece fits quite well, though it doesn’t breathe at all so it’s not great for wearing for all that long. It also has a habit of pinching my knuckle when I straighten my finger and the hinges can pinch when I flex my finger too. On the other hand, it flexes with my finger so well that if it were a little narrower it would be easy to forget I’m wearing it. The end is cut a little bit short so it doesn’t impede my fingertip and I can actually type while wearing it.

Version 3

Version 3 is going to again be a nearly fresh restart, largely to refactor the math to make the different parts share functions instead of being slightly modified copies of each other. The finger base piece turned out quite nicely in the end, but I’d rather not go through the hour+ I spent making the top actually flat again the next time I modify it.

I also have ideas on how to streamline the hinges so they’re more firmly attached and might not need supports.  So far all my supports have been part of the model.  I’ve tried letting Slic3r generate the supports for me but I’m printing directly on glass and the automatic support bases are so small that they get either knocked over or picked up and stuck on the side somewhere else.  My own supports have worked about half the time, which is much better but not exactly a great success rate.

Infill Patterns

I spent a decent bit of time yesterday tuning my printer for using the black PLA I ordered with it, and since I was printing ~50mm squares to go under the printer feet so it would sit flat on the wire shelf where my old printer was I had good top/bottom surfaces to play with alternate infill patterns:
Infill Patterns
Clockwise from top left (as labeled in slic3r): octogram spiral, hilbert curve, archimedean chords, concentric
Not shown: rectilinear (the default)

Of those I like the hilbert curve pattern the best, but slic3r isn’t kidding when it labels that with “(slow)” – it took a remarkably long time to do the top and bottom layers of that print. In general I’ll probably stick with rectilinear for most prints, but I may play with the others some more when I’m not printing objects with such large flat surfaces on top/bottom.

I probably should take a better picture of those using natural light, but that’ll be tricky now that I’ve finally gotten the sd card reader to work (card format matters: fat16 works) and put the printer where I intend to keep it:
Printer installed

I will need to figure out a lighting solution for it at some point: that corner has never needed light and keeping a flashlight nearby is somewhat less than ideal for examining prints-in-progress.

Talore was remarkably unbothered by having the printer there, though it looked like she was having trouble sleeping through the vibration of 40% honeycomb infill, which shakes the entire shelf a bit (but not nearly as much as it shakes my project table). I don’t think Dash has really discovered the printer yet. I would think he’d be interested in it when it’s moving, but I don’t know that he’s been high enough up to actually see on top of the build platform while it’s in motion yet (his hammock is a bit lower than that shelf).

Anyway, now that I’ve got the printer pretty well set for that filament I’m printing some small calibration objects to fine tune it and see if I can drop the layer height to .1mm (I have one passable print at .1mm with the blue ABS that I think I now know enough to improve, but my first try with PLA didn’t work at all).

Pile of Experience

I spent most of today trying to figure out how to get large prints in ABS to complete without curling up in the corners and ruining the print. Here’s the stack of LCD faceplate attempts that I produced.
Many Failures

The first attempt used the same settings as the calibration cube, but after about an hour it had curled enough in two corners that it was clear it wasn’t going to complete in a useful state. The second through fourth prints were attempts to print with the printer set to faster speeds to see if it would be able to keep ahead of the curl. Turned out no, and I had to bump the temperature up by a good bit to keep it from randomly jamming because it was trying to push plastic through the nozzle at about the same speed that the plastic was melting. The last prints I started playing with infill settings – it’s not a structural piece so the default of .4 infill is probably way overkill. I also had to slow down the travel time because it skipped steps a couple times.

The successful print was with .1 infill, a 1 mm brim, and the travel speed limited to 200 mm/s (down from a high of 500 mm/s). It still took over half an hour to print and there was slight curling in one corner (I’ll try increasing the brim next time to see if that’s actually what was helping). I think I’m going to need a proper heated chamber to print large pieces of ABS (luckily most of what I want to do in ABS is small pieces), so I might start trying to work with PLA tomorrow even though it’s not recommended for the E3d hotend I have.

The banding I was getting last night turned out to be from one of the threaded rods being bent. The design prevents it from pushing the extruder back and forth so I didn’t expect it to be a problem, but as it turned it was raising/lowering the end of the X-axis from where it should have been. I don’t think I quite have it perfect yet, but it’s much better. I also smoothed out the motion of the Z-axis while I had the rods removed so now I shouldn’t have any binding.

Also, because I recorded it, here’s a video of a short print from start to finish: